- What is unique about how research tasks are managed for FightAIDS@Home – Phase 2? How does this impact how work will be distributed to volunteers?
- What is trickle messaging?
- What is a trickle up message?
- Why is my device uploading result data (intermediate upload files) while a research task is still in progress?
- Why did my device stop processing a research task? And what is a trickle down message?
- Why is this project using intermediate uploads and trickle messages?
- Why does this application require zero redundancy, i.e. no comparison of results across devices?
What is unique about how research tasks are managed for FightAIDS@Home – Phase 2? How does this impact how work will be distributed to volunteers?
The simulations researchers need to carry out are typically very long running and complex, each of which would take several months to run on a single device. Instead, researchers split up each simulation for a given drug candidate into shorter running research tasks: much smaller and more manageable pieces. These pieces of work can be run independently and simultaneously on a volunteer devices.
However for FightAIDS@Home – Phase 2, the research tasks within a single drug candidate simulation are dependent on each other where the output of one task is used as the input to the next. This means longer research tasks within each drug candidate simulation which can’t be run simultaneously.
To handle this complexity, we are using two different, but related mechanisms called trickle messaging and intermediate uploads to allow us to track your progress through a research task and manage the handover of that task from one volunteer to the next to get it completed in the shortest time possible. This way, we can track the progress of the long simulations to ensure that computations are not delayed or lost, while the researchers get the valuable results back as quickly as possible. In addition, volunteers acquire their credits sooner too.
What is trickle messaging?
Trickle messaging is a capability that allows your device to communicate with us while working on a research task to report the progress being made by your device. This allows us to determine whether work should continue on that research task or whether insufficient progress is being made and therefore that task should be handed over to another volunteer for processing.
This capability is particularly useful to a project like FightAIDS@Home – Phase 2 because of the nature of its research tasks, which may require more processing time to complete and can therefore be started by one volunteer and completed by another, without losing the progress made by the first volunteer.
What is a trickle up message?
A trickle up message is when your device sends a message back to World Community Grid at certain processing milestones to inform us that you’re still making progress on the current research task. Along with intermediate results sent to us by your device, we use this information to:
- Validate your work up to that point and grant credit accordingly;
- Determine whether sufficient progress is being made by your device or whether the task should be handed over to another volunteer.
Why is my device uploading result data (intermediate upload files) while a research task is still in progress?
We are using a mechanism called intermediate uploads whereby at certain processing milestones, your device would send us back partial results for the research task your device is currently working on. This allows us to validate the work you have completed up to that point and helps the researchers examine and interpret the results being returned by the volunteers.
Why did my device stop processing a research task? And what is a trickle down message?
Using the information your device intermittently sends to us during the processing of a research task, we determine the likelihood of your device finishing the task before the completion deadline. If we determine that you are very likely to miss that deadline or you have already missed the deadline, we would send your device a trickle down message to instruct it to stop working on that research task and we then pass it along on to another volunteer.
There are two types of trickle down messages:
- Soft stop: Instruction for your device to continue until the next milestone before stopping the computation of the current research task. This happens when your device is not making sufficient progress on the current calculation. We would then hand over your partial result to another volunteer to continue working on.
- Hard stop: Instruction for your device to stop working on the current research task immediately. This happens when you have already passed the processing deadline without sending in the final full result or that there may be a communication error in sending us your progress. In this case, we would hand over the research task from the point of your last intermediate milestone to another volunteer to resume working on.
In either case, you will be awarded credit for the work you completed up to the point of the last checkpoint.
This mechanism allows work to be completed quicker and for the researchers to receive valuable results sooner.
Why is this project using intermediate uploads and trickle messages?
Unlike many of our research projects, FightAIDS@Home – Phase 2 requires that research tasks within each simulation, and small processing steps within each task, to be carried out in sequence. This means that it would take much longer to get the results to the researchers.
Trickle messaging and intermediate upload capabilities allow us instead to move the same research task from one volunteer to the next, without losing progress along the way. Overall, this assures progress, shortens and stabilizes the processing time required to complete research tasks and speeds up the rate at which valuable results can be returned to the researchers.
While a typical FightAIDS@Home – Phase 2 simulation might take up to a year to complete, using these capabilities means that it can be completed in as little as two months.
Why does this application require zero redundancy, i.e. no comparison of results across devices?
Unlike many of our research projects, FightAIDS@Home – Phase 2 does not require redundancy, where the same research task is sent to two devices and the results are compared for consistency. Instead, this project will be using various processing metrics during the computation of a research task to validate that the task is progressing without errors.